Please note this document is a working draft.

1 Introduction

1.1 Background

Cities have used urban green spaces as a tool to enhance the health of residents, increase biodiversity, and mitigate the effects of climate change. Research has shown that urban green space can alleviate urban-heat-island effects, improve air quality, and encourage outdoor socialization and physical activity.1 However, urban green spaces are more frequently supported in higher-income, predominantly white neighborhoods, exacerbating inequity in health.2 Several U.S. cities, including Denver, have proposed goals to equitably increase green space in their cities.

Often, urban planners and residents use turf and exotic plants to expand green space, which are are not ideal for pollinator health and demand high amounts of water, which may not be sustainable in the mountain west, a region expected to experience more severe drought due to climate change. While a growing body of research suggests exposure to urban green space can improve human health and prevent premature mortality,1,3 little is known about the potential impact of native vegetation on human health in cities in the U.S. mountain west, a region with a growing population vulnerable to the impacts of climate change.

1.2 Objective

We aim to estimate the number of premature deaths that would be prevented by the implementation of various policy scenarios adding green space, including native vegetation, to the City of Denver. These policy scenarios were informed by conversations with local stakeholders in advocacy, research, and governance.

2 Methods

2.1 Measurement of existing green space and definition of native-plants

We measured existing green space in the City of Denver by the normalized difference vegetation index (NDVI) measured by the Landsat-8 satellite at a spatial resolution of 30 square meters. To define native-plants greenness, we measured the NDVI of several places in Denver (places described in more detail here) comprised of various levels of native and adapted plants during cloud-free summer (May, June, July, August) days over five years (2016-2021). Over these days in these areas, the mean NDVI was as low as 0.35 at the Denver Botanic Gardens Green Roof and as high as 0.67 in City Park, which is 0% native and consists of more watered turf. A plot in Denver Botanic Gardens that is 100% native plants had a mean NDVI of 0.54 over these days.

Table 2.1: NDVI of selected places of varying levels of plant nativity on selected cloud-free summer (May, June, July, August) days over five years (2016-2021)
Place type Place name NDVI, mean NDVI, 25th-ile NDVI, 50th-ile NDVI, 75th-ile
high diversity Chatfield Meadow Restoration 0.48 0.47 0.50 0.50
high diversity Chatfield Prairie Garden 0.37 0.35 0.37 0.37
high diversity City Park Greenhouses 0.31 0.27 0.28 0.28
high diversity Denver Botanic Gardens Green Roof 0.34 0.34 0.36 0.36
high diversity Kenderick Lake Xeriscape Garden 0.40 0.36 0.37 0.37
native spectrum Denver Botanic Gardens, 100% Native 0.54 0.52 0.55 0.55
native spectrum Plains Conservation Center, 100% Native 0.16 0.23 0.26 0.26
native spectrum Green Mountain Park, 85% Native 0.42 0.36 0.40 0.40
native spectrum Hogback along C-470, 75% Native 0.36 0.32 0.35 0.35
native spectrum Suburban Open Space 1, Chatfield H.S., 50% Native 0.40 0.37 0.41 0.41
native spectrum Suburban Open Space 2, Columbine Hills Church, 30% Native 0.36 0.26 0.31 0.31
native spectrum Denver Botanic Gardens at Chatfield, 10% Native 0.53 0.40 0.54 0.54
native spectrum City Park, 0% Native 0.67 0.67 0.70 0.70

To acknowledge the lack of precision in these NDVI estimates and the range of NDVI levels that may correspond to a native-plants intervention, we defined NDVI using the mean of the Denver Botanic Gardens as as the primary analysis and that of the Denver Botanic Gardens Green Roof as an alternative value.

2.2 Stakeholder engagement to develop policy scenarios

We conducted ten interviews with local stakeholders who are actively engaged in advocacy, policy, and governance related to native plants and green space in Greater Denver. Following these interviews, we developed four policy scenarios informed by the collective set of visions and priorities articulated by the stakeholders. The scenarios range from short-term goals to more ambitious, perhaps less immediately realistic visions. The scenarios fall into five categories.

  1. Homogenous native-plants greening: Under the first set of scenarios, we consider adding native plants homogenously across all block groups, without respect to the specific location. Specifically, we consider two scenarios:

    • Set 30% of the block group’s area to be as green as native plants, corresponding with a goal set by some scientists and advocates to protect 30% of lands and oceans by 20304;

    • Set 20% of the block group’s area to be as green as native plants, a level that, according to some of our interviewees is sufficient to support pollinator health.

  2. Equity-driven native plants greening: Under the second set of scenarios, we propose adding native plants to the same amount of area as above (30% or 20% of Denver’s area) but rather than applying the intervention homogeneously across all black groups, we prioritize application of the native plants intervention to those areas of greatest need from a health-equity perspective. Both the State of Colorado and the City of Denver have created equity indices to guide their work. The Colorado Department of Public Health & Environment (CDPHE) describes their equity index here5 and makes the index available for download here.6 The City of Denver Department of Public Health & Environment (DDPHE) describes the neighborhood-level equity index here7 and makes the data available for download here.8 We use both definitions to consider a total of four scenarios:

    • Set 30% of Denver’s area to be as green as native plants, and apply this intervention to those block groups most in need based on the CDPHE definition of health equity.

    • Set 20% of Denver’s area to be as green as native plants, and apply this intervention to those block groups most in need based on the CDPHE definition of health equity.

    • Set 30% of Denver’s area to be as green as native plants, and apply this intervention to those block groups most in need based on the DDPHE definition of health equity.

    • Set 20% of Denver’s area to be as green as native plants, and apply this intervention to those block groups most in need based on the DDPHE definition of health equity.

  3. Create native buffers around riparian areas (rivers, streams, lakes, and ponds) of the following sizes:

    • 200 feet (ideal for ecosystem health but possibly not realistic)

    • 100 feet (compromise)

    • 50 feet (most realistic; less good for ecosystem health)

  4. Initiatives related to green infrastructure and stormwater management. (Please refer to corresponding section for details.)

  5. Replace a portion of parking-lot surface with native plants:

    • 100% of the existing parking area (unrealistic, out of curiosity)
    • 50% of the existing parking area (perhaps realistic)
    • 20% of the existing parking area (most realistic)

2.3 Visualization of scenarios

This section elaborates on the scenarios and visualizes them with maps.

2.3.1 Scenario 1: add native plants to all census block groups

Under the first scenario, which we view as the most ambitious, we consider the expected impact on mortality by applying a green-space intervention to all census block groups, without specifying where.

We first measured the mean NDVI of each census block group on July 4, 2021. The weighted block-group-level mean is presented here, weighted by the proportion each 30 square-meter pixel covers by the census block group. For example, if half of a pixel overlaps the block group, it receives a weight of 0.5 in the weighted average. We also removed bodies of water before measuring NDVI.

Figure 2.1: Scenario 1

For each definition of native-plants NDVI, We excluded census blocks whose baseline NDVI was above the threshold. This map visualizes the NDVI cutoffs both for lower of the two values, that of the Denver Botanic Gardens Green Roof, and the higher of the two, that of Denver Botanic Gardens Native Plot.

2.3.2 Scenario 2: add native plants prioritizing census block groups with most health-equity need

Under this set of scenarios, we assumed we had a “budget” of greening of either 30% or 20% of Denver’s area, and we applied that budget to those with most need from a health-equity standpoint based on the CDPHE and DDPHE definitions, described above. As the below maps show, the two health-equity definitions from the State and from the City overlap considerably.

2.3.2.1 30% of area prioritizing block groups of greatest need according to two health-equity definitions

2.3.2.2 20% of area prioritizing block groups of greatest need according to two health-equity definitions

2.3.3 Scenario 3: add native-plants buffers to riparian areas

We measured NDVI in a 200-foot buffer, a 100-foot buffer, and a 50-foot buffer around all bodies of water in Denver. We downloaded bodies of water from OpenStreetMap (code here). We defined residential exposure to these riparian areas as those individuals living within a 500-meter buffer, following research on green-space and health that have defined green-space exposure based on residential proximity.3 We estimated the number of people in this buffer by multiplying the population density of the census block group by the intersecting area.

The below map depicts mean NDVI in the portions of census block groups that intersect a 200-foot buffer as well as those pieces that intersect the part of the 500 m buffer that would not be intervened upon, i.e., the part between the 200-foot buffer and the edge of the 500 m buffer.

2.3.4 Scenario 4: green infrastructure

We spoke with representatives at the local Office of Green Infrastructure who described three categories of initiatives–some planned, some aspirational–that could add native or adapted plants to the city.

  • Large stormwater retention projects:
    • Large ponds or basin located on public property that treat collect and treat stormwater after it has been collected in a storm pipe. They are usually vegetated, often with native or adapted plants.
    • Our colleague expected about 75% of the facility’s footprint would have native or adapted plants.
  • Green streets; according to our colleague:
    • Historically, about 2.7 miles of green streets each year, and each mile equates to about 0.15 acres of native or vegetated landscape.
    • Short term goal: increase output to 5.0 miles of green streets; same amount of vegetation per green mile
    • Aspirational goal: output to 5.0 miles; increase vegetated area to 0.75 acres per green mile
  • Proposed stormwater controls on new or re-development:
    • Our colleague stated that as properties develop or redevelop, they may be required to include stormwater runoff control measures to offset negative impacts to flooding and water quality downstream of the site associated with the impervious surfaces added during the development. These stormwater control measures are often green-on-the-ground practices vegetated with native or adapted plants.
    • The requirements may differ by parcel size. Our colleague estimated the following number of sites that may require stormwater control by parcel size:
      • greater than 1 acre: about 100 sites per year
      • 0.5-1.0 acre: about 25 sites per year
      • less than 0.5 acre: 400 sites per year

2.3.4.1 4.1. Large stormwater retention ponds

We were provided a list of planned projects throughout Denver. Per our conversations, we assume that about 75% of the project’s footprint would consist of native or vegetation. Like for riparian areas, we defined exposure to the projects as those individuals living within a 500 m buffer of the projects. A map of baseline NDVI of the projects themselves and of a 500 m buffer around the projects is below.

2.3.4.2 4.2. Green streets

Not yet complete.

2.3.4.3 4.3. Site new or re-development storwmater controls

The approach for estimating the health impact of the redevelopment controls differs from scenarios 2 and 3.1. because we will not know where the re-developments will occur. We thus simulated possible locations. Per our conversations, we anticipate the following number of parcels will be subject to these rules per year:

  • greater than 1 acre: about 100 sites per year
  • 0.5-1.0 acre: about 25 sites per year
  • less than 0.5 acre: 400 sites per year

We gathered data on parcels from Denver’s Open Data Portal (Existing Landuse 2018) and measured their area. A subset of these parcels near Union Station is mapped below.

Then, from all parcels, we sampled 100 sites of size greater than 1 acre, 25 sites of size 0.5-1 acre, and 400 sites less than 0.5 acres. One such sample appears below.

From this point, we followed the same framework as for other scenarios. The baseline NDVI of the sampled parcels and that of their corresponding 500m buffers is visualized below.

2.3.5 Scenario 5: add native plants to parking lots

Finally, we propose replacing a portion of parking-lot surface with native plants. Denver’s area is about 9% parking lot, so even replacing a small amount of parking-lot surface could have a large impact on the total area of urban greening. We obtained spatial data on parking lots](https://www.denvergov.org/opendata/dataset/parking-lots-2016)) from the City of Denver Open Data Catalog.

Total area (mi2), parking lots Total area (mi2), Denver proportion parking lots
13.57 155 0.09

We similarly measured the NDVI on the parking lots as well as the NDVI in the census block groups within a 500-m buffer radius of any parking lot (which is almost the whole city). We present a small subset of these areas near Union Station:

2.4 Health-impact assessment methods

We estimated the number of premature deaths averted under each scenario by following a recent meta-analysis that estimated that for every 0.1 unit increase in exposure to NDVI, the relative risk of premature death decreases about 4% (pooled risk ratio of 0.96; 95% confidence interval [CI]: 0.94, 0.97).3 We used this risk ratio to estimate the population attributable fraction9 corresponding to the proposed changes in NDVI from the baseline levels for each scenario to the alternative native-plants level.

Under the first two scenarios, the areal unit is the block group itself, and because we do not specify where, exactly, in the block group the greening intervention would occur, we define NDVI exposure as the average NDVI in the block group and compare that value with the alternative native-plants value. We use 5-year American Community Survey Data (2016-2020) to estimate the population in 5-year age groups in each census block group. For the other scenarios, we draw a 500-m buffer around the intervention area, as shown above. To estimate the population affected, we multiply the population density of the block group in that age group by the area of the block group covered by the intervention area.

To estimate the number of premature deaths prevented in each age group and area, we multiply the population-attributable fraction by the baseline mortality rate by the population size. We restrict analyses to adults aged 30 and above (subject to change) following the age range of many of the cohort studies reviewed3 and exclude northeastern census block groups near the airport (subject to change; perhaps not for all scenarios).

We estimate uncertainty in both the block-group-level population estimates and the relative risk estimate (95% CI: 0.94, 0.97) by resampling from normal distributions and taking the 2.5th and 97.5th percentiles over 1,000 replications. This code has additional detail.

3 Results

Results of each scenario appear in the table below. The first table describes, for each scenario, the population affected, the cumulative area of the treatment itself (e.g., the area of the retention ponds or parking lots) and the cumulative area of the 500 m residential buffer surrounding the treatment area. Note for Scenarios 1 and 2, the treatment area and the residential-buffer area are the same (the area of the affected census block groups) because we do not say exactly where in the block groups the greening treatment would occur. # Tables ## Table 1: Description of scenarios (area, pop affected, NDVi)

Table 3.1: For each scenario and native-plants NDVI definition, the cumulative area of the treatment, its corresponding residential buffer, and the population affected, overall and stratified by equity tertile.
Scenario NDVI, native Equity Tertile Area, treatment (mi2) Area, residential buffer (mi2) Baseline NDVI Alternative NDVI Pop. affected, est. Pop. affected, 95% LL Pop. affected, 95% UL
all-bg 0.34 NA 1,161 1,161 0.23 0.27 152,144 147,510 156,534
all-bg 0.54 NA 2,619 2,619 0.34 0.40 385,872 383,180 393,183
all-bg 0.34 [0,32.4] 350 350 0.24 0.27 40,941 38,577 43,431
all-bg 0.34 (32.4,55.3] 252 252 0.23 0.27 50,241 47,883 52,112
all-bg 0.34 (55.3,91.1] 559 559 0.23 0.26 60,962 59,432 63,626
all-bg 0.54 [0,32.4] 1,036 1,036 0.37 0.42 143,548 141,565 146,066
all-bg 0.54 (32.4,55.3] 725 725 0.36 0.41 124,985 122,339 128,492
all-bg 0.54 (55.3,91.1] 858 858 0.28 0.36 117,339 115,562 121,321
riparian 0.34 NA 72 641 0.22 0.23 70,043 67,956 72,377
riparian 0.54 NA 163 1,364 0.33 0.35 182,229 179,015 186,351
riparian 0.34 [0,32.4] 29 237 0.24 0.24 22,835 21,195 24,152
riparian 0.34 (32.4,55.3] 15 141 0.22 0.23 22,057 20,894 22,749
riparian 0.34 (55.3,91.1] 28 263 0.20 0.21 25,151 23,692 26,511
riparian 0.54 [0,32.4] 68 543 0.36 0.37 64,140 62,981 66,254
riparian 0.54 (32.4,55.3] 45 371 0.35 0.36 58,748 57,025 60,745
riparian 0.54 (55.3,91.1] 50 450 0.28 0.30 59,342 57,723 60,763
ogi 0.34 NA 11 261 0.21 0.21 45,386 43,813 47,101
ogi 0.54 NA 17 442 0.29 0.29 75,217 72,804 77,698
ogi 0.34 [0,32.4] 1 35 0.20 0.20 7,845 6,914 8,529
ogi 0.34 (32.4,55.3] 3 73 0.19 0.20 18,711 17,652 19,818
ogi 0.34 (55.3,91.1] 6 153 0.23 0.23 18,831 18,052 20,060
ogi 0.54 [0,32.4] 3 75 0.32 0.32 13,701 12,765 14,779
ogi 0.54 (32.4,55.3] 4 104 0.25 0.26 23,636 22,708 25,583
ogi 0.54 (55.3,91.1] 10 263 0.29 0.30 37,881 36,106 39,235
prkng 0.34 NA 18 1,166 0.23 0.24 152,955 149,454 154,998
prkng 0.54 NA 25 2,585 0.34 0.35 379,132 373,338 386,792
prkng 0.34 [0,32.4] 5 351 0.24 0.25 41,821 40,185 43,781
prkng 0.34 (32.4,55.3] 4 252 0.23 0.24 49,948 48,030 51,414
prkng 0.34 (55.3,91.1] 10 562 0.22 0.24 61,186 58,649 63,269
prkng 0.54 [0,32.4] 8 1,013 0.37 0.38 140,532 137,799 143,843
prkng 0.54 (32.4,55.3] 6 719 0.36 0.37 122,759 120,642 125,233
prkng 0.54 (55.3,91.1] 11 854 0.28 0.29 115,840 111,233 121,964

3.1 Table 2: estimated deaths averted (total and rate per 100k) under each scenario

##  [1] "scenario"                      "scenario_sub"                 
##  [3] "scenario_sort_order"           "ndvi_native_threshold"        
##  [5] "pop_affected"                  "area_mi2_bg_int_tx"           
##  [7] "area_mi2_bg_int_res"           "ndvi_mean_alt_int"            
##  [9] "ndvi_quo_int"                  "ndvi_mean_alt"                
## [11] "ndvi_quo"                      "ndvi_diff"                    
## [13] "deaths_prevented"              "deaths_prevented_per_pop"     
## [15] "deaths_prevented_per_pop_100k" "attrib_d"
Table 3.2: For each scenario and native-plants NDVI definition, the cumulative area of the treatment, its corresponding residential buffer, and the population affected, overall and stratified by equity tertile.
Scenario NDVI, native Equity Tertile NDVI linear change Deaths prevented, est. Deaths prevented, 95% LL Deaths prevented, 95% UL Deaths prevented per 100k, est. Deaths prevented per 100k, 95% LL Deaths prevented per 100k, 95% UL
all-bg 0.34 NA 0.03 15 13 18 10 9 12
all-bg 0.54 NA 0.06 88 85 93 23 22 24
all-bg 0.34 [0,32.4] 0.03 3 2 3 7 6 8
all-bg 0.34 (32.4,55.3] 0.03 6 4 9 12 8 18
all-bg 0.34 (55.3,91.1] 0.04 6 5 7 10 9 11
all-bg 0.54 [0,32.4] 0.05 25 24 28 18 17 19
all-bg 0.54 (32.4,55.3] 0.05 32 29 34 25 23 27
all-bg 0.54 (55.3,91.1] 0.08 31 29 34 27 25 28
riparian 0.34 NA 0.01 1 1 2 2 1 2
riparian 0.54 NA 0.02 12 10 13 6 5 7
riparian 0.34 [0,32.4] 0.00 0 0 1 2 1 4
riparian 0.34 (32.4,55.3] 0.01 0 0 0 1 1 2
riparian 0.34 (55.3,91.1] 0.01 0 0 1 1 1 2
riparian 0.54 [0,32.4] 0.02 4 3 5 6 5 7
riparian 0.54 (32.4,55.3] 0.02 4 3 5 6 5 8
riparian 0.54 (55.3,91.1] 0.02 4 3 5 7 6 8
ogi 0.34 NA 0.00 0 0 1 1 0 2
ogi 0.54 NA 0.00 2 1 2 2 1 3
ogi 0.34 [0,32.4] 0.00 0 0 0 1 1 2
ogi 0.34 (32.4,55.3] 0.00 0 0 1 2 0 4
ogi 0.34 (55.3,91.1] 0.00 0 0 0 0 -1 0
ogi 0.54 [0,32.4] 0.00 0 0 0 2 1 2
ogi 0.54 (32.4,55.3] 0.01 1 0 1 4 2 5
ogi 0.54 (55.3,91.1] 0.00 0 0 0 1 1 1
prkng 0.34 NA 0.01 4 4 5 3 3 3
prkng 0.54 NA 0.01 14 13 15 4 3 4
prkng 0.34 [0,32.4] 0.01 1 1 1 2 2 2
prkng 0.34 (32.4,55.3] 0.01 2 1 2 3 2 4
prkng 0.34 (55.3,91.1] 0.01 2 2 2 3 3 4
prkng 0.54 [0,32.4] 0.01 4 3 4 3 2 3
prkng 0.54 (32.4,55.3] 0.01 6 4 6 4 4 5
prkng 0.54 (55.3,91.1] 0.02 5 4 6 4 4 5

4 Next steps

As of April 27, 2022:

  • Add green streets HIA

  • Add economic-impact attributable to health

References

1.
2.
Wen M, Zhang X, Harris CD, Holt JB, Croft JB. Spatial Disparities in the Distribution of Parks and Green Spaces in the USA. Annals of Behavioral Medicine. 2013;45(S1):18-27. doi:10.1007/s12160-012-9426-x
3.
Rojas-Rueda D, Nieuwenhuijsen MJ, Gascon M, Perez-Leon D, Mudu P. Green spaces and mortality: A systematic review and meta-analysis of cohort studies. The Lancet Planetary Health. 2019;3(11):e469-e477. doi:10.1016/S2542-5196(19)30215-3
4.
Dinerstein E, Vynne C, Sala E, et al. A Global Deal For Nature: Guiding principles, milestones, and targets. Science Advances. 2019;5(4):eaaw2869. doi:10.1126/sciadv.aaw2869
5.
CDPHE. Climate Equity Data Viewer. Published online September 13, 2021. https://storymaps.arcgis.com/stories/46bf289f92bc4629a0a1266de4bb7f97
6.
7.
8.
Denver. Denver open data catalog: Equity index 2020 - neighborhood. https://www.denvergov.org/opendata/dataset/city-and-county-of-denver-equity-index-2020-neighborhood
9.
Murray CJ, Ezzati M, Lopez AD, Rodgers A, Vander Hoorn S. Comparative quantification of health risks: Conceptual framework and methodological issues. Population Health Metrics. 2003;1(1):1. doi:10.1186/1478-7954-1-1


Copyright © 2022 Michael D. Garber